Reminds me of a college project/paper I did back in the day.
Fully kitted out a Rabbit with airdam, nice sloping grille block, wipers "up" and antenna remove, mirrors folded in, and rear wheel skirts. Found the flattest, emptiest piece of road I could with the calmest conditions I could wait for and made several runs timed with stopwatch.
Wanted in the worst way to report improvements!!!
Buuuuutttt... presented the numbers as they were and if they weren't lying, there was no statistical improvement.
Could be my designs weren't good. Well they weren't all that unique; they followed convention. Does this sort of thing need lots of work to be optimalized?
Could be my testing wasn't accurate enough. But then, if the improvement was that small an increment, what good is it?